The present invention relates generally to infrastructure resource provisioning, and more particularly to infrastructure resource provisioning using trace-based workload temporal analysis for high performance computing.
High performance computing applications are being moved from traditional high performance computing centers to infrastructure (or cloud) providers. Typically, system administrators who want to setup an environment similar to their on-premises infrastructure counterpart have no tooling available that assist in this task. The system administrators need to pick resources at the Infrastructure as a Service (IaaS) layer and adjust such resources over time. If the resources picked is underutilized, this process may generate resource waste. If not enough resources are provisioned, this process may result in bad Quality of Service (QoS) for users.